In partnership with
On 17 May 2026, a drone struck the perimeter of Barakah. It hit a generator outside the inner fence. There were no injuries and no radioactive release. The reactors never stopped. The Gulf's largest power base absorbed an attack and held.
Barakah supplies 5.6 gigawatts (GW). Its four reactors deliver about 25% of the United Arab Emirates' electricity. One in four Emirati electrons is now nuclear. That base is built, proven, and resilient.
The artificial intelligence (AI) it could feed is not built yet. Gulf generation runs ahead of Gulf compute. This inverts the Western pattern, where demand outruns power. It opens a financing window few outside the region have priced.
US interconnection queues hold 2,600 GW. Barakah already delivers 5.6 GW. The Gulf built the power first.
Western infrastructure starts with AI demand. Operators sign power deals, then wait years for electrons. In the United States, interconnection queues hold thousands of gigawatts. Power is the bottleneck.
The Gulf reversed the order. The UAE finished Barakah first. A KEPCO-led consortium built four APR-1400 reactors in roughly a decade. Output: 40 terawatt-hours (TWh) a year. Westinghouse chased Gulf reactor work for years without landing the job. Delivery certainty won the trust that underwrites the compute phase.
Saudi Arabia shows the same ordering. Its sovereign AI champion, Humain, launched in May 2025 under the Public Investment Fund (PIF). The generation to serve it is planned alongside, not before.
The catch is timing. Reactors take a decade. Data halls take 18 months. The slow asset is already finished. The fast one has barely started.
Merchant equity funds compute, then chases power. Three financing models. All three struggle here.
Western financing assumes power follows demand. Three structures dominate, and each struggles here.
First, merchant equity. It funds compute, then chases power, rewarding speed over patient generation. Nuclear's decade-long build repels it.
Second, the power purchase agreement (PPA). A hyperscaler signs a long offtake to de-risk a plant. Few will sign for reactors not yet licensed.
Third, project finance. Lenders want predictable cash flows and completion guarantees. Commissioning risk frightens them, so capital arrives late.
Each model treats power as the variable and compute as the constant. The Gulf treats power as the constant. One finances a promise. The other finances a fact.
Microsoft pledged £11.2bn to the UAE. G42 owns the AI stack. The state owns the power beneath it.
Sovereign structures change the economics. The capital is patient, concentrated, and aligned with the state's industrial aim.
MGX, launched in 2024 by G42 and Mubadala, invests across the AI stack. Abu Dhabi shows the template. Microsoft committed £1.1bn ($1.5bn, €1.3bn) to G42 in 2024. It then pledged £11.2bn ($15.2bn, €13bn) across the UAE to 2029. A UAE-US AI campus is planned at 5 GW. First phase: 1 GW.
Saudi Arabia concentrates capital through PIF. Humain has agreed a financing framework for AI infrastructure worth up to £880m ($1.2bn, €1bn). It aims for 6 GW by 2034 and wants to be the world's third-largest AI provider.
Qatar adds a third model. Ooredoo launched a sovereign AI cloud on Nvidia hardware. Backing: around £740m ($1bn, €850m).
These are not subsidies. They are equity stakes by states that already own the power. Sovereign equity carries no offtake clause and no completion bond. It accepts a decade horizon. The state captures the industrial return.
In your market, which is further ahead, firm power or AI compute?
Reactors take a decade. Data halls take 18 months. In the UAE, the slow asset finished first.
Market observers track the AI race at the top of the stack, counting chips and data halls. The binding constraint sits lower, at generation and grid. The Gulf understood this first.
The arbitrage is temporal. A reactor started today reaches the grid near 2035. A data hall started today opens in 2027. In most markets, the reactor is the laggard. In the UAE, it already runs. The compute can be built against power that exists.
That sequence compresses risk. Across the GCC, more than £22bn ($30bn, €26bn) is committed to AI data centres by 2030. The 17 May strike tested the base. The grid held. Resilient power lowers the risk premium on everything built downstream.
The UAE built a nuclear regulator before it built a reactor. That credibility now underwrites the AI phase.
The Federal Authority for Nuclear Regulation (FANR) licensed Barakah under international review. That institutional credibility now extends to the compute phase.
Saudi Arabia is moving from policy to licensing. It has stood up its own regulator and engaged the International Atomic Energy Agency (IAEA). Engineers and regulators, not algorithms, sign off each reactor. The GCC Interconnection Authority has linked six national grids since 2009. Bahrain, with no reactors, draws on that shared network.
Follow the electrons. Build compute where firm power already runs, not where it is promised.
Three principles guide capital into this market.
Three principles guide capital into this market.
Match the capital to the asset. Patient sovereign equity suits decade-long generation. Merchant money does not.
Price the sequence. Where energy precedes load, commissioning risk falls. For operators, that means siting near Barakah's firm output. Behind-the-meter load avoids grid queues entirely.
The Gulf did not win the AI race by building the smartest models. It won the first round by building the power.
Barakah proves the base is real, resilient, and already feeding a grid. The compute will follow the electrons. That is the only direction the physics allows. Energy is not downstream of AI. It is the precondition.
Next week: The commissioning hedge, and why operators should insure against 18 months of dark.
Stop babysitting dashboards. Ship from Slack. Touch grass.
700+ teams have Viktor reading their Google Ads every morning.
Your media team opens Slack at 8am. There's a cross-platform brief in #growth: Google Ads spend vs. ROAS, Meta CPA by campaign, Stripe revenue by channel. Viktor posted it at 6am. Nobody asked for it.
Last week, one team's Viktor caught a spend spike at 2am on a broad match campaign and flagged it in Slack: "CPA up 340%. Recommend pausing and shifting budget to the top two performers." That would have burned $3K by morning. The media buyer woke up to a problem already handled.
Your strategist reviews spend trends. Your account manager checks revenue attribution. Same Slack channel, same colleague, before anyone's first coffee.
Google Ads, Meta, Stripe. One message. No Looker, no Data Studio. Anomaly detection runs around the clock. Cross-platform reporting runs on autopilot.
5,700+ teams. SOC 2 certified. Your data never trains models.
"Viktor is now an integral team member, and after weeks of use we still feel we haven't uncovered the full potential." — Patrick O'Doherty, Director, Yarra Web